Immediately follows the compressed XML metainfo, of size XMLCompSize. First dword of it is the size of uncompressed data, the rest is zlib compressed unicode XML.
If version>=800, the gif thumbnail follows, of size Gif Size.
Object index
Offset to the index is specified by the Object Index Offset in the header, and number of entries is Number Of Objects.
Root Object ID field in the LRF header points to a Book Atr object, which sets some common properties for the book, and in turn points to a Page Tree object. Page Tree has a list of Pages in the book, in the reading order. A Page can contains some Blocks and Canvases. Each Block can be positioned on the page indepentently. A Block can contain a single passage of Text, Simple Text, Image, Button (and some others, I think). Text and Simple Text contain a stream of text with possibly some embedded tags. An Image points to an Image Stream object, which contains the binary stream of the image. (Several Images can point to a common Image Stream, each representing a portion of the whole bitmap.) Button can be used as a hyperlink – to jump to another portion of the book.
Toc Object ID and Toc Object Offset in the header point to a single Toc (table of contents) object, which is basically a list of links to some parts of the book.
http://groups.yahoo.com/group/librie/message/153